The scientific planning of medical care today requires improvements in existing data and methods for studying the clinical course of chronic disease. In cancer and in various degenerative diseases, the classification of patients does not include many critical features needed to estimate prognosis, and to evaluate therapy. Although the clinical course and statistics for management of a particular disease can be greatly affected by the patient's pattern of symptoms, associated ailments, and decisions about seeking and accepting medical assistance, these clinical, chronometric, co-morbid and decisional data are seldom carefully assembled, classified, and correlated in the appraisal of therapeutic accomplishment. Because the intellectual "landmarks" have not been clearly specified and analyzed, investigators encounter major difficulty and controversy in evaluating the merits of different programs of medical care. Our object in this research is to distinguish these "landmarks" by assembling and analyzing appropriate data; by constructing improved taxonomic systems for classifying and storing the data; by clinically integrating computers into the management of the data and by developing better methods for predicting prognosis, evaluating therapy, and designing trials of future new methods of therapeutic care. In preliminary studies of medical record data for patients with cancers of the lung, rectum, and larynx, we have already demonstrated the feasibility, practicality, and value of these new approaches. Our object now is to complete the final analyses used for creating new systems of pre-therapeutic staging, for developing both automated and judgmental procedures to improve the process of multivariate prognostic stratification, and for applying these new techniques in evaluating the accomplishments of therapy.